Forced Turnover:
Evaluating Pressing Effectiveness in Soccer

Natalie Rayce, David Almona

Definitions

Pressing: a defensive tactic where players apply coordinated pressure on the opponent with the ball to force mistakes, win back possession, and quickly transition to attack

Forced Turnover: when a player loses possession due to opponent pressure, resulting in the opposing team gaining control. This includes misplaced passes, interceptions, successful tackles, or losing control under pressure - all direct results of effective defensive pressure

Data Overview

  • Dataset: 521 matches in the MLS 2023 season
  • Source: SkillCorner
  • Three data types:
    • Match information: game details (teams, pitch, referee)
    • Event data: player actions (passes, shots, tackles) with timestamps and coordinates
    • Tracking data: real-time positions of all players and ball (10 Hz)

Methods

  • Standardized the tracking data, such that home team always attacks left-to-right
  • Calculated frame-by-frame distance, speed, acceleration, and direction for every player on the field
  • Identified the ball carrier for each frame using the events data and measured how fast defenders were moving toward the ball carrier

Methods: Press detection

Our pressing criteria required defenders to meet two conditions:

  • Be within 6 meters of the ball carrier, AND
  • Either be approaching at more than 1.0 meters per second OR show close engagement behaviors

The Modeling Data

  • 4,875 pressing sequences identified across the first 10 MLS matches
  • 27 features capturing tactical, spatial, and temporal contexts
  • Target Variable: Forced turnover within 5 seconds of press sequence start (binary)

Key Variables

  • Spatial Context: Ball carrier position, distance to boundaries, field third
  • Pressing Dynamics: Number of defenders, approach velocity, passing options
  • Game Context: Score, game state (winning/losing/drawing), time remaining
  • Situational Factors: How ball carrier gained possession (pass reception, interception, etc.), incoming pass characteristics (distance, height, range)

Results

  • Model: XGBoost, with 10-fold cross validation
  • Accuracy: 82.1%
  • ROC-AUC: 0.782

Confusion Matrix:

FALSE TRUE
FALSE 883 172
TRUE 20 138

Results: Variable Importance

Results: ROC Curve

Limitations

  • Inaccuracy in SkillCorner data
  • Individual player skill and tactical tendencies not considered in the mode

Next Steps

  1. Use an elliptical (or oval) pressure zone rather than a circular one.

Next Steps

  1. Compare the expected pressure difference from individuals and teams on a scatterplot.

Next Steps

  1. Now that we have an XGBoost model, we plan to compare it with other models, e.g random forest, lightGBM.

Appendix

?